Accelerometers find a wide variety of applications within modern motor vehicles. The most common of these are impact and collision sensors used to deploy front and side impact air bags in modern passenger cars and trucks.
In applications that depend on sudden and drastic deceleration, the presence of gravity is of little consequence and will not affect the implementation of the accelerometer. However, increasingly feedback systems within motor vehicles have attempted to make use of accelerometer data during much lower and subtler levels of acceleration.
One example is anti-collision warning systems. Though all street legal motor vehicles have brake lamps configured to signal other drivers of braking, these signals do not warn following drivers of imminent braking. At least one system has proposed activating a vehicle's brake lamp system in response to a deceleration signal from a sensitive accelerometer, and independent of actuation of the brake pedal. The system described in U.S. Pat. No. 6,411,204 to Bloomfield et al., entitled “DECELERATION BASED ANTI-COLLISION SAFETY LIGHT CONTROL FOR VEHICLE,” includes a plurality of deceleration thresholds each with an associated modulation of the brake lamps.
However, the system fails to precisely account for gravitational forces, limiting its effectiveness to deceleration regimes where gravity's effect is minimal and reducing its effectiveness as an early warning system. Accelerometers, known as tilt sensors in the gaming and robotics industries, are extremely sensitive to any gravitational force to which they are not perpendicular. This sensitivity complicates any system that attempts to detect low levels of acceleration by using accelerometers within moving vehicles, since the system must account for the wide variety of orientations of the accelerometer relative to the earth's gravity introduced as the vehicle travels uphill, downhill, through cambered or off-camber curves, and on cambered grades. For instance, an accelerometer in a vehicle stopped on a 45-degree downhill slope would sense deceleration of a magnitude equal to 0.71 times the acceleration due to gravity. To avoid gravitational acceleration artifacts, the system of Bloomfield only produces output if the deceleration signal rises above a predetermined threshold set above the level of artifacts introduced during typical driving conditions.
However, the reliance of this device on a threshold deceleration reduces its effectiveness as an early warning system. Even a short delay between the time when the subject vehicle begins to slow down and the time when a following vehicle begins to slow can result in a rapid closure of the gap, or following distance, between the vehicles, and a potential collision. Consequently, the shorter the following distance between vehicles, the smaller the margin of error will be for drivers of following, vehicles to avoid rear-end collisions. Disengaging the accelerator, or coasting, is often the first response of the driver of a subject vehicle to observing a non-urgent traffic event in the roadway ahead, and usually results in a slight deceleration. By failing to warn other drivers of the possible imminence of braking of a subject vehicle, the proposed device loses valuable time. To avoid this problem, the thresh old must be set lower, which could result in gravitational acceleration artifacts affecting the system's output. For example, an overly low threshold could prevent the device from signaling deceleration on an uphill grade since the accelerometer would sense a component of the earth's gravity as acceleration. Similarly, a low threshold could cause the device to continuously flash during a descent, while gravity appears as deceleration.
The loss of time incurred by a threshold-based system might be tolerable in some other application; but in collision prevention, even an instant saved can prevent a collision. A Special Investigative Report issued in January of 2001 by the National Transportation Safety Board (NTSB) illustrates the scale of the problem. The report notes that in 1999 “1.848 Million rear-end collisions on US roads kill[ed] thousands and injur[ed] approximately [one] Million people.” The report concluded that even a slightly earlier warning could prevent many rear-end collisions.                Regardless of the individual circumstances, the drivers in these accidents were unable to detect slowed or stopped traffic and to stop their vehicles in time to prevent a rear-end collision. If passenger car drivers have a 0.5-second additional warning time, about 60 percent of rear-end collisions can be prevented. An extra second of warning time can prevent about 90 percent of rear-end collisions. [NTSB Special Investigative Report SIR—January 2001, Vehicle- and Infrastructure-based Technology for the Prevention of Rear-end Collisions]        